We compared the sensitivity of different types of visual analog scales and of descriptive pain terms in healthy volunteers and in postoperative patients. One hundred and seven volunteers marked visual analog scales according to their perception of the descriptive pain terms--little, mild, some, moderate, severe, agonizing. Individual variation in values and preferences between the five following five different visual analog scales were analyzed: 10-cm linear horizontal and vertical scales, a curvilinear scale, and graded horizontal and curvilinear scales. Significantly more volunteers preferred the horizontal scale with gradations. Scores for the vertical linear scale had the greatest coefficient of variation and were least normally distributed. The majority of volunteers considered the phrase "agonizing pain" the best term defining the extreme limit of pain (X2(12) = 41.8, P less than 0.001). There were significant changes in the values of pain intensity measured on visual analog scales by patients using the same descriptive pain term on successive observations. However, the patients' values for pain terms in the preoperative pain-free state were not significantly different from those during postoperative pain. We conclude that graded linear horizontal scales are both more reliable and preferred by participants and that visual analog scales give a more sensitive and accurate representation of pain intensity than do descriptive pain scales.